Analysis · curated 30 Sep 2026
OWASP Top 10 for Agentic Applications
First reported aikido.dev
Coverage timeline
Single-source analysis — first reported, latest, and curated coincide.
Why it matters
The OWASP Top 10 for Agentic Applications gives defenders a structured checklist of the main risk classes—prompt injection, memory poisoning, insecure tool use, and RAG poisoning—to test when deploying copilots and tool-using LLM agents.
Aikido's documentation page enumerates an OWASP Top 10 for Agentic Applications, describing the agentic AI risk classes it tests during pentests, including prompt injection, sensitive information disclosure, excessive agency, insecure tool use, insecure output handling, memory poisoning, retrieval/knowledge-base poisoning, and authentication/authorization failures. The content is reference material describing risk categories rather than any specific vulnerability or event.